Exploratory Causal Analysis for Networks of Ethnographically Well-studied Populations Sociological Methodology and Research

Annotations added. This draft needs a comparison of results for ordinal probit and o^t linearizing power parameter for the models. We need to include the Male mortality in War variables, and include a State/Nonstate variable (v237: JurisHier) I. Introduction to estimation methods for societal databases. This study is concerned with better understanding of how to evaluate networks of causal relationships that are identified and connected – and as a prerequisite modeled through regression models that are well specified and cross-validated – in ways adequate for Bayesian causal models. One feature of such models, for example, is not to violate well-established counterfactuals. Counterfactuals are among the many criteria that allow linkage of results of regression analyses into networks of potential causes and effects. Many other criteria must be satisfied as well to evaluate causal relationships (e.g., Pearl 1987, 2000, 2009b), however, some of which are explored here. We begin with common problems of identification for datasets distinguished by levels of aggregation (e.g., individual vs. societal), in this case, the study of networks of ethnographically well-studied populations. (1) For purpose of comparison with earlier studies, our focus of analysis is on